Spatial Variability of Sampling: Grid Size Impact on Atmospheric Metals and Trace Elements Deposition Mapping with Mosses

This paper deals with the spatial heterogeneity structure and the influence of the precision of moss data on results interpretation as real spatial variations of atmospheric heavy metal deposition. Different sizes of map mesh net (unit I of 30 × 30 km, unit II of 10 × 10 km, unit III of 3 × 3 km) have divided an area of 90 × 90 km. The protocol used is a nested design with three levels and random draws. The statistical method of components of variance analysis estimates the associated variability for each mesh size. Our results show the poor precision of this biological tool for map purposes on a little scale (unit III). Furthermore, the high residual variance of As, Ba, Cd, Co, Cu, La, Ti, and U hides the spatial variations associated with mesh sizes. In order to obtain useful maps, it should be reasonable to use a 30 × 30 km mesh size, or even larger, to build spatial variation maps of Pb, Sb and with more caution for Cu, Sr, Rb and Zn. For V, As and Cd, the residual variability of moss data was too important to guarantee any spatial origin to the mapped variations. The cost–benefit study shows that the sampling effort has to be concentrated on unit I of 30 × 30 km to optimize future campaigns, and with a particular stress on the sampling repetitions for Cu, Pb, and Sb.